The homicide rate in Brazil has been fairly stable during the past two decades when compared to other Latin American countries like Colombia and Mexico, fluctuating around an alarming 28 deaths per 100,000 population. However, the stability at the national level hides a drastic change in the spatial distribution of violence within the country. Between 2000 and 2017, the homicide rate almost halved in the Southeast of Brazil, the most violent region throughout the 90's, while increasing by 145 percent in the North and the Northeast. Some researchers speculate that this spatial shift of homicides may be related to demographic factors, crack epidemics, changes in police behavior or different enforcement of federal laws. Nevertheless, there is no solid research that discusses the causes of this shift to the best of our knowledge. We hypothesize a different mechanism that could explain this shift — the dynamics of criminal organizations’ activities and the market competition among them and its consequences for homicides. Anecdotal evidence suggests that the dynamics of criminal activity related to illegal drug trade is able to explain this shift in violence’ patterns. Yet, the evidence about criminal dynamics is mostly qualitative and provided in a decentralized manner from few states and cities, curbing scholars’ ability to study criminal violence in a more rigorous and data-driven approach. We intend to overcome the lack of data availability by constructing a novel data of spatial and temporal presence of criminal organizations in Brazil since 2000 and test the hypothesis that the dynamics of criminal organizations is responsible for these patterns.